Transport Research Laboratory Creating the future of transport

Transcription

1 Transport Research Laboratory Creating the future of transport PUBLISHED PROJECT REPORT PPR704 TfL Cycle Facility Trials: Alternative Separation Methods for Cycle Lanes G Beard Prepared for: Project Ref: Transport for London M12 Alternative Separations Quality approved: S Greenshields (Project Manager) M Jones (Technical Referee) Transport Research Laboratory 2014

2 Disclaimer This report has been produced by the Transport Research Laboratory under a contract with Transport for London. Any views expressed in this report are not necessarily those of Transport for London. The information contained herein is the property of TRL Limited and does not necessarily reflect the views or policies of the customer for whom this report was prepared. Whilst every effort has been made to ensure that the matter presented in this report is relevant, accurate and up-to-date, TRL Limited cannot accept any liability for any error or omission, or reliance on part or all of the content in another context. When purchased in hard copy, this publication is printed on paper that is FSC (Forest Stewardship Council) and TCF (Totally Chlorine Free) registered. Contents amendment record This report has been amended and issued as follows: Version Date Description Editor Technical Referee V1 19/08/14 Final version GB MJ

3 Contents 1 Introduction 7 2 Cyclists Introduction What impact did the method of cycle lane separation have on cyclist usability? Ratings of understanding the cycle lane markings Ratings of ease of using the cycle lane What impact did the method of cycle lane separation have on cyclist safety? Ratings of perceived safety Lateral position of cyclists within the cycle lane What impact did vehicle interaction have on cyclist usability? Effect of vehicle interaction on ratings of ease of using the cycle lane What impact did the vehicle interaction have on cyclist safety? Effect of vehicle interaction on ratings of perceived safety Effect of vehicle interaction on lateral position of cyclists within cycle lane Further comments Summary and conclusion 25 3 Other road users Introduction What impact did the method of cycle lane separation have on driver usability? Ratings of understanding the cycle lane markings Ratings of ease of navigating the trial route What impact did the method of cycle lane separation have on driver perceptions of safety? Ratings of perceived safety Lateral position of vehicles within lane Further comments Summary and conclusion 40 4 Pedestrians Introduction April PPR704

4 4.2 What impact did the method of cycle lane separation have on pedestrian usability? What impact did the method of cycle lane separation have on pedestrian safety? Further comments Summary and conclusion 44 5 Summary and conclusion Cyclist behaviour Driver behaviour Pedestrian behaviour Implications 46 6 References 49 April PPR704

5 Executive summary Introduction Transport for London (TfL) commissioned the Transport Research Laboratory (TRL) to conduct a series of trials to examine the impact of different methods of cycle lane separation on the behaviour and safety of road users (inc. cyclists, car drivers, motorcyclists, HGV drivers and pedestrians). The methods of separation investigated were: 1. a kerb with 365 mm hard margin (full continuous segregation with physical barrier); 2. bolt-on delineators: the Zicla Zebra 9 (a type of intermittent separation with low-profile barriers positioned in 2.5 m intervals); 3. 1-m high marker posts: Jislon wands (intermittent separation with high-profile barriers positioned in 2.0 m intervals) In addition, a painted solid white line (mandatory cycle lane line, continuous separation with no physical barrier) was used to provide an experimental baseline, enabling researchers to control for differences between different groups of participants on different trial days. Objectives Cyclist and driver trials aimed to determine the extent to which the method of cycle lane separation influenced: the behaviour (in particular speed and position in the lane); and perceptions of road users, with a focus on usability (i.e. understanding of the cycle lane markings and ease of navigating the route) and perceptions of safety. Pedestrian trials aimed to determine the extent to which the method of cycle lane separation influenced the perceptions of pedestrians when crossing the road. Method Each separation method was trialled on a separate day, so each trial day represents a different sample of participants. Each group of participants experienced the white line separation on one section of the TRL test track and only one out of the three physical separations on a different part of the track. For each day, therefore, comparisons were made between the white line separation and the method of physical separation. To directly compare the three types of physical separation tested on different days (i.e. the treatment conditions), difference scores were calculated relative to the ratings of the white line separation (i.e. the control condition): Difference score = Treatment score Control score Thus, a positive difference indicates that the physical separation (the treatment) received higher scores than the white line separation (the control). The conclusions April PPR704

6 summarised below and throughout this report are related to difference scores for each of the three physical separation methods. Key findings Key statistically significant findings 1 have been reported so as to provide direct conclusions relating to each road user group. In general, cyclist, driver and pedestrian ratings of perceived usability and safety were high for all separations (wands, Zebras, kerb and white line) indicating that all four methods were viewed as fairly safe and fairly easy to use. Nonetheless, statistically significant differences were found between the methods, indicating small but distinct variations in the perceptions and behaviour of the different road users. Such variations may have design implications when implementing integrated and segregated cycle lanes. Low profile intermittent separation: Zicla Zebras Of the three physical separations investigated, intermittent separation provided by the Zebras was found to offer the smallest improvements to the perceived usability and safety of cyclists and car drivers. For motorcyclists and HGV drivers, the Zebras (unlike other forms of physical segregation) were perceived to be less safe and harder to navigate past than a painted white line. For example, ratings of safety for the Zebras were 4-5% lower for motorcyclists and 15% lower for HGV drivers compared to ratings for the white line. This observation is consistent with qualitative comments reported by motorcyclists indicating they considered themselves at greater risk of collision with the Zebra separation. Full continuous separation: Kerb with 365-mm hard-margin The use of a solid kerb was preferred over a painted white line by cyclists and car drivers, but not by motorcyclists or pedestrians. For example, riding past the kerb separation was rated about 8% more difficult and 5-8% less safe than riding past the white line separation by motorcyclists. Evidence from motorcyclists chosen riding path supports the conclusion that they were more aware of a risk of injury from collision with the kerb, thereby increasing the cognitive load of the riding task and reducing their perceptions of safety. When calculated against a standardised baseline, the lateral position of motorcyclists indicated that they travelled, on average, around 230 mm further from the edge of the kerb than from the edge of Zebras and around 110 mm further than from the edge of the wand separation. Due to perceived increased difficulty of crossing the road and reduced safety when crossing a road with a kerb separated cycle lane, the kerb was also rated as least favourable for pedestrians compared to the white line separation. Difference scores related to the perceived safety of cyclists were larger for the kerb (mean, M = 0.75) than the Zebras (M = 0.25), indicating that perceived safety was greater with the kerb. 1 Statistically significant indicates any pattern or relationship in the data that has a probability of occurring by chance of less than 5% (i.e. p < 0.05). April PPR704

7 Consistent with this finding, cyclists travelled an average of approximately 100 mm closer to the kerb than to the Zebras; this supports evidence from questionnaire ratings showing an increased perception of safety when protected by a kerb. High profile intermittent separation: Jislon wands The Jislon wands were the only physical separation method which offered improved perceptions of safety and usability over white line separation for all road users (except pedestrians where no significant differences were identified). Difference scores related to cyclists perceived safety of wand separation (M = 0.54) were significantly larger than those for intermittent Zebra separation (M = 0.25), but were not significantly different than those for kerb separation (M = 0.75). This indicates that perceived safety was greater with the wands. Cyclists travelled closer to the wand separation than to the other separation methods. Calculation of lateral position against a standardised baseline indicated cyclists passed on average up to 280 mm further away from the Zebras and around 100 mm further away from the kerb separation. The lateral position of vehicles showed that car drivers, motorcyclists and HGV drivers allowed a clearance from the wand separation around 230 mm, 120 mm and 170 mm greater, respectively, than from the Zebras. Motorcyclists also travelled around 110 mm closer to the wands than the kerb. Conclusions Comparison of the alternative methods of cycle lane separation revealed statistically significant differences in the speed and road position of cyclists and drivers, and in their perceptions of usability and safety. However, the extent of those differences was not so large as to raise fundamental objections to the use of any one method. Cyclists rode in fairly central positions in the lane with all separation methods, and at similar speeds. Importantly though, the width of the separation methods differs substantially, ranging from 100 mm wide for the wands to 365 mm for the kerb. The latter would normally be 500 mm wide so as to provide a physical buffer between cyclists and vehicles with large wing-mirrors (such as HGVs and buses). There are clearly great savings in the road space required to install the intermittent methods. However, implications of reducing the width of this buffer must be considered, especially given the observation that cyclists rode closer to the wands (suggesting they made greater use of the available width). Of the separations investigated, continuous kerb separation and intermittent wand separation may offer the greatest benefits to cyclists. Whilst intermittent Zebra separation offers some benefits for the safety and usability of cyclists compared to a painted white line, the extent of those benefits is smaller than with other physical separations. The wands were the only physical separation method which showed increases in the perceived safety and usability of motorcyclists and HGV drivers compared to the white line separation. Zebra and kerb separation was found to decrease the ease of navigating the route and reduce feelings of safety for motorcyclists. Given the limitations of any offstreet trial, it cannot be assumed that the same findings would be replicated in a real street environment, so further study through on-street trials is needed before definitive design recommendations can be made. April PPR704

8

9 1 Introduction Traditional methods for segregating cyclists from traffic have tended to involve either taking cyclists off the carriageway completely, sharing space with pedestrians (which can lead to loss of priority as well as other difficulties), or significant infrastructure construction for methods such as kerb segregation within the carriageway. This requires a lot of road space, which can limit the choice of locations where it can be installed, as well as having significant implications for drainage, maintenance, road sweeping and gritting etc. For these reasons there has been recent interest in so-called light methods for separating cyclists from the traffic, which utilise a variety of different forms of intermittent features (such as blocks, planters, or bolt-on delineators), vertical barriers, or wands (such as bollards or marker posts), or enhancement of conventional lane markings (such as rumble strips and reflective studs). Light methods require less space, and are therefore able to be more widely used, without interfering with drainage and also, by virtue of not creating a continuous barrier, enable cyclists to leave the separated cycle lane should they need to do so (for example, when turning right at junctions). However, consideration must be given to how cyclists and other road users will respond to such facilities, for example, how do light methods affect perceptions of comfort and safety, and what impacts do they have on the chosen position of road users in the lane? As part of a wider programme of off-street trials of innovative cycling infrastructure, the Transport Research Laboratory (TRL) was commissioned by Transport for London (TfL) to examine the impact of different methods of cycle lane separation (see Table 1) on the behaviour and safety of a) cyclists, b) other road users (i.e. car drivers, motorcyclists and HGV drivers), and c) pedestrians. Following an initial review of literature on separation methods, which was mostly international guidance, it was decided to carry out off-street trials of two examples of intermittent separation, one high-profile and one low. The objective was to compare these with full segregation (via a solid kerb) and a white-line only mandatory cycle lane, representing traditional approaches to separation. While variants on these basic approaches also exist, such as using reflective studs and textured markings, it would be expected that user responses would lie within the range of the methods tested. The key research questions examined in these trials were: What impact did the method of cycle lane separation have on usability? What impact did the method of cycle lane separation have on safety? It was also of interest to understand the influence of the presence of other road users (i.e. the vehicle interaction) on cyclist behaviour and safety, specifically: What impact did the vehicle interaction have on cyclist usability? What impact did the vehicle interaction have on cyclist safety? Each of the alternative methods of cycle lane separation were presented to participant cyclists, car drivers, motorcyclists and HGV drivers who were asked to travel along a set route on the TRL test track. The route was split into two sections of road; one with a mandatory cycle lane marked with a painted white line (i.e. the control condition) and one with a cycle lane separated by one of the three physical methods (i.e. a kerb, wands, or Zebras). Each participant took part in one trial day only, so they all experienced the white line separation and only one of the three physical separations. Whilst the method of physical separation was varied between the trial days, the white April PPR704

10 line separation remained unchanged across all trial days, thereby enabling a consistent reference on which to base all comparisons between the three physical methods. This was crucial to the design of the trial, necessary to control for confounding factors relating to variability between the different groups of participants who took part on different trials days. Hence, when comparisons are made between the three methods of physical separation, this is done by comparing results from three different sets of participants, each using the same section of track; while comparisons between white line separation and each physical separation involves comparing results from the same group of participants, but on different sections of track. Table 1 Methods of separation tested in trials. Type of separation Method used Notation used in this report Type of trial variable Continuous separation with no physical barrier Solid painted white line white line Control Continuous segregation with a physical barrier Kerb with 365 mm 2 hard margin kerb Experimental/ Treatment Intermittent separation with low profile barriers Zicla Zebra 9 bolt-on delineators (a.k.a. armadillos ) Zebra Experimental/ Treatment Intermittent separation with high profile barriers 1-m high marker posts: Jislon wands wand Experimental/ Treatment The main objective of the trials was to determine the extent to which the method of cycle lane separation influenced the behaviour (in particular speed and position in the lane) and perceptions of road users, with a focus on usability (i.e. understanding of the cycle lanes and ease of navigating the route) and feelings of safety. Data for the three 2 Standard specification for solid kerb segregation indicates a hard-margin of 500 mm, however this width could not be achieved for the trials due to space constraints on the test track. April PPR704

11 physical separation methods (i.e. the treatment conditions) are presented relative to data obtained for the white line separation (i.e. the control condition). The influence of specific interactions between vehicles and cyclists at the entrance, the middle and the exit of the cycle lanes was also examined. Data were provided by questionnaires administered on-track during the trial and offtrack after the trial, and by traffic counters located on the trial route. The Appendices describe the data handling and statistical analysis used for this study. Statistical analysis of the questionnaire and traffic counter data has enabled identification of findings that are statistically significant (i.e. any pattern or relationship in the data that has a small probability of occurring by chance). It is commonly accepted that if a finding has occurred with a confidence level of 95% or more, then it did not occur by chance (i.e., p < 0.05), and it is statistically significant. Sometimes the probability of a chance finding will be less than 5% and this is expressed accordingly (e.g. p < means probability was less than 0.1%). Full results from the statistical analysis are available in the Appendices. Analysis of qualitative reports given by participants (e.g. for some questions in the offtrack questionnaire participants were asked to give free text answers), focused on identification of common themes in responses based on the subject of discussion, the type of separation method or the type of road user. The purpose of this report is to present the key statistically significant findings and common qualitative themes from the trials so as to provide direct answers to the four research questions. Non-statistically significant results have not been discussed. Section 2 deals with results related to cyclists, section 3 presents the results related to other road users and section 4 presents the findings from the pedestrian trial. The results are summarised and implications are discussed in section 5. Information regarding the participant sample, the research methodology, example questionnaires used for data collection, photographs of the trial sites and full details of statistical analysis used for this study and the associated statistical output can be found in the separate Appendices document. April PPR704

12

13 2 Cyclists 2.1 Introduction This section presents the findings from the cyclist behaviour trials, which investigated the influence of the method of cycle lane separation on the user behaviour and safety of 241 participant cyclists. The influence of specific interactions with cars whilst using the cycle lanes was also examined. Usability (i.e. understanding of the cycle lanes and ease of navigating the route) and perceptions of safety were assessed via several means. During the trial, an on-track questionnaire was used to obtain ratings related to the comprehension of cycle lane markings, the ease of using the cycle lane, and perceptions of safety. These topics were also further assessed using an off-track questionnaire after each trial session. Traffic survey tubes were used to record the speed and lateral position of cyclists using the cycle lanes under different traffic conditions. This section collates data from all three of these sources in order to establish key findings for the cyclist user group. Comparisons have been drawn between the physical and non-physical separation methods (i.e. white line vs. wands/zebras/kerb) and between each of the three physical separation methods (i.e. wands vs. Zebras vs. kerb). 2.2 What impact did the method of cycle lane separation have on cyclist usability? Ratings of understanding the cycle lane markings Physical separation vs. no physical separation In this section, and elsewhere in the report, the term cycle lane markings is used to refer to the structure and composition of the cycle lane separation (e.g. the presence of a solid kerb) and is not strictly limited to painted markings or signage. Whilst the cyclists were on-track, they were asked to rate the ease of understanding the cycle lane markings after their first encounter with the white line separation and the physical separation. Cyclists responses revealed no significant differences between the white line separation and any of the physical separation methods (i.e. wands, Zebras, or kerb), suggesting their first impressions of the markings were comparable between physical and non-physical separation. An off-track questionnaire was used to probe further into cyclists views on various aspects of the cycle lane markings, including how clearly the markings identified the cycle lane and conveyed understanding of how to proceed, and how easy or difficult it was to understand where the cycle lane began and finished. Some significant differences in perception of the markings were identified from the off-track questionnaire. In general, mean ratings were quite high (> 3, where 5 is the top of the scale indicating very clear, or very easy to understand), indicating cyclists found it fairly easy to understand the markings. However, compared to non-physical separation (i.e. the white line), statistical analysis revealed that: April PPR704

14 Markings on the cycle lane separated by a kerb resulted in clearer identification of the cycle lane and more clearly showed how to proceed onto and through the cycle lane. Cyclists found it significantly easier to understand where the cycle lane began and finished with the wand separation than with the white line. It was easier to understand where the cycle lane finished with the Zebras than with the white line. Overall, responses to the off-track questionnaire show that the markings on the cycle lanes separated by physical methods (i.e. wands, Ziclas and a kerb) were clearer and easier to understand than those on the cycle lane separated by a painted white line. It is particularly clear that cyclists found it easier to understand where the cycle lane finished for all physical methods of separation compared to the cycle lane separated by only a white line, with the greatest improvements in understanding found for the wand and kerb separation methods. These results suggest that the use of physical separation methods may help to clearly differentiate the cycle lane from the main carriageway Effect of the type of physical separation Each separation method was trialled on a separate day, meaning each trial day represents a different sample of participants. Each group of participants therefore experienced the white line section and only one out of the three physical separation methods. To directly compare ratings of the cycle lane markings across each of the types of physical separations (i.e. across different days) scores were calculated relative to ratings of the markings on the white line separation (a.k.a. difference scores, see Equation 1). Equation 1: Difference score =Treatment score - Control score where the Treatment Scores were ratings for the physical separation methods (i.e. wands, Zebras, kerb) and the Control Scores were ratings for the non-physical separation (i.e. the white line). Ratings from the on-track questionnaire obtained cyclists first impressions and analysis showed no statistically significant differences in the difference scores between the cycle lane separated by wands, Zebras, or a kerb. The off-track questionnaire allowed participants to make a more reflective assessment of the cycle lane markings. Mean difference scores for the off-track questionnaire ratings of the cycle lane markings are shown in Figure 1. As the difference scores are all positive it is clear that cyclists rated all three physical methods higher than the white line, scoring on average 0.04 to 0.45 higher than the white line. Statistical analysis revealed the following statistically significant findings: Ratings of ease of identification of the cycle lane markings were higher for the kerb separation (mean difference score, M = 0.23) than the Zebra separation (M = 0.07) and the wand separation (M = 0.04) Markings at the entrance and the exit were easier to understand on the cycle lane separated by wands (M = 0.36 and 0.45, respectively) than equivalent markings with Zebra separation (M = 0.10 and 0.17, respectively). April PPR704

15 Markings at the entrance were rated higher for the wands (M = 0.36) than the kerb (M = 0.05). Wands Zebras Kerb Mean difference scores Identification of markings Proceeding through cycle lane Markings at entrance Markings at exit Question topic in off-track questionnaire Figure 1 Mean difference scores for off-track questionnaire ratings of cycle lane markings for each method of separation (positive difference indicates physical method rated higher than white line). On-track questionnaire ratings obtained cyclists first impressions whilst off-track questionnaire ratings allowed a more reflective assessment. Significant differences were only found for off-track questionnaire ratings, indicating that the method of physical separation influenced the perception of the cycle lane markings, but that there may be little impact on behaviour when actually using the cycle lanes, at least for the first time. Overall though, results indicate that: The continuous barrier of the kerb separation enabled easier identification of the cycle lane than either the Zebras or wands, but: Ratings of ease of using the cycle lane Physical separation vs. no physical separation During the trial cyclists were asked to rate the ease of using the cycle lane with the white line separation and the physical separation. Analysis of these ratings revealed that physically separated cycle lanes were, in general, easier to use than the cycle lane separated by a white line. Specifically, the wand separation and the kerb separation were rated as significantly easier to use than the cycle lane separated by a white line, but no significant differences were found between the Zebras and the white line. From the off-track questionnaire, the wand separation was also rated as easier to use (with one car present) than the cycle lane separated by a white line. Likewise, the kerb April PPR704

16 separation was rated as easier to use (with one car present and in scenarios with busy town centre traffic) than the white line separation. Overall, the cycling task was easier when the cycle lane was separated by either 1m high wands or a solid kerb with a 365 mm hard margin than when separated by a painted white line. No differences in the ease of use were found between the Zebras and the white line separation. The improvement in usability associated with kerb separation was particularly evident when respondents were asked to imagine using the cycle lane in busy town centre traffic Effect of the type of physical separation As with the analysis of cycle lane markings, direct comparison between ratings of the ease of using the cycle lanes for each of the physical separation methods was performed by calculating scores relative to the ratings of the white line separation (see Equation 1, above). This was necessary because each separation method was trialled on a separate day with different groups of participants. Statistical analysis revealed no significant differences in on-track questionnaire ratings of ease of using the cycle lane between any of the physical separation methods. Mean difference scores for the off-track questionnaire ratings are shown in Figure 2. As the difference scores are all positive it is clear that cyclists considered all three physical methods to be easier to use than the white line, scoring on average 0.01 to 0.51 higher than the white line. Statistical analysis revealed significantly higher difference scores for ratings of ease of using the cycle lane (in conditions of busy traffic) for the kerb separation (M = 0.51) than both the wand separation (M = 0.1) and the Zebra separation (M = 0.07) Wands Zebras Kerb Mean difference scores Ease of use with no traffic Ease of use with 1 car present Question topic in off-track questionnaire Ease of use in busy traffic Figure 2 Mean difference scores for off-track questionnaire ratings of ease of using the cycle lane for each method of separation (positive difference indicates physical method rated higher than white line). April PPR704

17 Overall then, there is some evidence to suggest that cycling in conditions of busy traffic was considered an easier task when using a cycle lane continuously separated by a solid kerb than by other intermittent separation methods (as indicated by responses to the off-track questionnaire) when respondents were asked to imagine using the lane in busy traffic Cyclist speed The speed of cyclists (m/s) when using each of the physical separation methods (wands, Zebras, kerb) was compared using the white line separation as a baseline measure, as shown in Equation 2: Equation 2: Speed difference = Treatment speed - Mean control speed where the treatment speeds were individual speeds (m/s) measured on the physically separated cycle lanes (i.e. wands, Zebras, kerb) and the mean control speeds were the mean speeds (m/s) measured on the non-physically separated cycle lane (i.e. the white line). Negative speed differences indicate that the speed of cyclists using the physically separated cycle lanes was lower than when using the cycle lane separated by a white line; almost all speed differences were negative as the white line cycle lane was positioned on the same downhill slope and the physically separated lanes on the same uphill slope. However, it is the relative differences between the three methods which are important, as the slopes were the same for all trial days. Figure 3 shows the mean differences in cyclist speed for the entrance, the middle and the exit of each cycle lane (the mean speed difference measured at the middle of the kerb separation is not shown due to missing traffic counter data). The differences in speed are greatest at the exit of the cycle lanes; likely because this location was at the bottom of a slope on the control lane and at the top of a slope on the treatment lane. 0 Wands Zebras Kerb Mean speed differences (m/s) Entrance Middle Exit Location on cycle lane Figure 3 Mean speed differences for each physical method of separation (Positive differences indicate speed was greater with physical separations than the white line, negative differences indicate the opposite). April PPR704

18 Statistical analysis revealed small but statistically significant effects of the type of physical separation on cyclist speed at all three locations on the cycle lane. The differences in the speed of cyclists shown in Figure 3 may be equated to absolute speeds by normalising the means relative to a baseline. Table 2 shows these speed measurements under the assumption that this baseline is equal to 7 (i.e. a speed of 7 m/s when using the white line separation). A value of 7 was selected for the baseline on the grounds that it is comparable to the mean speed of cyclists using the white line cycle lane. Relative to this baseline, cyclists were, on average: mph faster (depending on the measurement location, i.e. entrance, middle, exit) when using the kerb separation than when using the Zebra separation, and; mph faster (depending on the measurement location, i.e. entrance, middle, exit) when using the kerb separation than when using the wand separation. As shown in Table 2, cyclists were slowest when using the Zebra separation, although this was only slightly slower than when using the wands. Table 2 Mean absolute speed (ms -1 / mph) of cyclists in each cycle lane, normalised to a mean speed of 7 ms -1 on the white line cycle lane. Measurement location Normalised absolute speed (ms -1 / mph) [White line (i.e. 7 ms -1 ) + mean difference] White line Wands Zebras Kerb Entrance 7 / / / / Middle 7 / / / Exit 7 / / / / Small but statistically significant differences in cyclist speed were observed, with cyclists travelling faster when using the kerb separation than when using the other physical separation methods, when normalised to the speed measured on the white line separation. This may be explained by the increased ease of using the cycle lane separated by a kerb, as indicated by the questionnaire ratings of usability (see section 2.2.2). Evidence from research in driving suggests that people self-pace their behaviour in order to maintain a tolerable level of task difficulty. For example, drivers are known to reduce their speed to compensate for increases in perceived task difficulty (Fuller, 2000; 2008; 2011, as cited by Kinnear and Helman, 2013). The evidence from greater usability ratings supports the conclusion that greater confidence when using the kerb separation may have encouraged cyclists to travel through the lane at a faster speed compared to the other methods. April PPR704

19 2.3 What impact did the method of cycle lane separation have on cyclist safety? Ratings of perceived safety Physical separation vs. no physical separation Perceptions of safety were principally examined using the on-track questionnaire which asked participants to provide a rating on a 1 to 10 scale, where 1 = very unsafe and 10 = very safe. Figure 4 shows the mean on-track safety ratings for the white line separation and each of the physical methods of separation. In general, while mean safety scores were high (> 8.8) for all methods, cyclists felt safer when using physical separation rather than non-physical separation. Statistically significant differences in on-track questionnaire ratings showed that cyclists felt safer when using the cycle lane separated by wands, Zebras and a kerb than with the white line. The greatest improvements in safety were found with the kerb separation (which showed about an 8% increase in mean ratings relative to the white line separation). Mean ratings of safety (on-track questionnaire) White line Wands White line Ziclas White line Kerb Day 1 Day 2 Day 3 Trial day / Method of separation Figure 4 Mean on-track questionnaire ratings of safety for each method of separation (where 1 = very unsafe and 10 = very safe). Perceptions of safety were also examined using the off-track questionnaire. The wand separation and the kerb separation were consistently rated as more safe than the white line separation when: One car was present in the road, and; Imagining using the lane in busy town centre traffic. Likewise, the Zebra separation was rated as safer than the white line when using the lane with one car present in the road. April PPR704

20 Physical separation between the cycle lane and the main carriageway improves cyclists perceptions of safety compared to non-physical separation Effect of type of physical separation As each separation method was trialled on a separate day, it is important to remember that each trial day represents a different sample of participants. Each group of participants therefore experienced the white line section and only one out of the three physical separation methods. For each day, therefore, Figure 4 (above) provides a comparison between the perceived safety scores for white lines and the method of physical separation, but comparisons should not be made between days (i.e. between physical separation methods), since it cannot be assumed that different groups of participants would have rated each separation method in a similar fashion. Overall, it is clear that participants felt very safe on all days and with all separation methods, with the lowest mean score of 8.85 from the on-track questionnaire data (i.e. nearly at the top of the scale, where 10 indicates they felt very safe). To directly compare ratings of safety across each of the types of physical separation methods, scores were calculated relative to the ratings of the white line separation (see Equation 1). The mean difference scores for the wands, Zebras and kerb are shown in Figure Mean difference scores for on-track questionnaire ratings of safety Wands Zebras Kerb Method of separation Figure 5 Mean difference scores for on-track questionnaire ratings of safety for each method of separation (positive differences indicate physical method rated safer than white line, negative differences indicate the opposite). As the difference scores are all positive it is clear that the participants considered all three physical separations to be safer than the white line, scoring, on average, between 0.25 and 0.75 points higher on the 10-point scale than the white line. Statistical analysis of these difference scores enabled assessment of which of the physical methods were perceived to be safest overall, on the basis of the difference from the control (i.e. the white line). April PPR704

21 Statistically significant effects of the type of physical cycle lane separation on perceived safety were identified. Specifically, analysis of the on-track safety ratings revealed: Both the wand separation (M = 0.54) and kerb separation (M = 0.75) were perceived to be safer than the Zebra separation (M = 0.25). No significant differences were found between the wand and kerb separation methods. This is illustrated in Figure 5 which shows that the difference scores for the wand and the kerb were similar. Using the same method of calculating difference scores, a significant effect of the method of physical separation on safety ratings was also found with the off-track questionnaire responses: The wand separation (M = 0.45) was rated as safer than the Zebra separation (M = 0.10) in conditions of busy traffic. The kerb separation (M = 0.54) was rated as safer than the Zebras (M = 0.19) when considering scenarios with one car present The kerb separation (M = 0.70) was rated as safer than the Zebras (M = 0.10) and in scenarios with busy traffic. Overall it can be concluded that: Cyclists felt very safe with all methods of separation, and; All the physical separation methods showed perceived safety benefits over white line. however; Cyclists considered the 1-m high wands and the 365 mm hard-margin kerb to offer a better safety barrier between the cycle lane and the main carriageway than the bolt-on Zebras Lateral position of cyclists within the cycle lane The lateral position of cyclists within the cycle lanes (i.e. the distance between the lefthand edge of the cycle lane separation and the bicycle wheel, in metres) was measured using traffic counters. As stated in section 1, comparisons between the three methods of physical separation have been made by comparing results from three different sets of participants, each using the same section of track; while comparisons between white line separation and each physical separation involves comparing results from the same group of participants using different sections of track. Here, the lateral position of cyclists in each physically separated cycle lane has been compared across the different groups of participants by normalising to a common reference (i.e. the white line cycle lane). Specifically, lateral position was calculated by subtracting the mean position measurement (in each trial session) on the white line cycle lane from individual position measurements on the physically separated cycle lanes (see Equation 3). Equation 3: Position difference = Treatment position - Mean control position where the treatment positions were individual measurements of lateral position (m) on the physically separated cycle lanes (i.e. wands, Zebras, kerb) and the mean control positions were the mean lateral position (m) on the non-physically separated cycle lane April PPR704

22 (i.e. the white line) measured during the same trial session. A positive difference in lateral position indicates the cyclist was further away from the separation method when using the physical separation than when using the white line separation, and a negative difference indicates the opposite. Statistical analysis revealed a significant overall effect of separation method on cyclist lateral position at all measurement locations. At all locations along the cycle lane, cyclists travelled closer to the wand separation than the Zebras or a solid kerb (except at the entrance to the cycle lane where cyclists were closer to the kerb than the wands). Cyclists also travelled closer to the kerb than the Zebras at the entrance and exit of the cycle lane. The differences in lateral position (calculated via Equation 3) may be reverted to absolute positions of cyclists in the lane by normalising the means relative to a single baseline measurement. Table 3 shows these position measurements under the assumption that this baseline is equal to 1 (i.e. a distance of 1 metre from the edge of the white line separation method). A value of 1 was selected for the baseline on the grounds that it is a reasonable approximation of the overall mean position of cyclists using the white line cycle lane, across all trial sessions. Table 3 Mean absolute lateral position (m) of cyclists in each cycle lane (distance from cycle lane separation), normalised to a mean position of 1 metre on the white line cycle lane. Measurement location Normalised absolute position (m) [White line (i.e. 1m) + mean difference] White line Wands Zebras Kerb Entrance Middle Exit Mean absolute lateral position of cyclists (m), normalised to 1m Wands Zebras Kerb NEARSIDE KERB CYCLE LANE CYCLE LANE SEPARATION ENTRANCE MIDDLE EXIT Measurement location Figure 6 Mean absolute position (m) of cyclists in each physically separated cycle lane (distance from separation), normalised to a mean position of 1 metre on the white line cycle lane. April PPR704

23 Figure 6 shows that cyclists generally travelled in fairly central positions, but that they tended to ride furthest from the Zebras and closest to the wands. This observation is consistent with evidence from the reduction in perceived safety observed with the Zebra separation compared to the wand separation (as shown by the questionnaire ratings discussed in section 2.3.1). Inexperienced or less confident cyclists frequently believe that the safest place to cycle is close to the left-hand edge of the road, as they may perceive this to minimise risk of conflict with passing vehicles. As such, cycle training programmes, for example Bikeability 3, specifically advise cyclists to ride further out in the lane to give themselves more space to avoid potholes, etc., and to discourage following drivers from overtaking where there is insufficient space. Hence, cyclists' willingness to ride further away from the left-hand road edge with the wand separation is consistent with greater perceived safety and confidence to occupy the road space. The observation that cyclists ride further out in lanes separated with wands means that they are using a greater proportion of the road space that has been re-allocated to them with this method of separation, further implying that the wands may be helpful in creating separated lanes where space is more constrained. This finding can be expressed by the separation efficiency ratios shown in Table 4, which is simply the ratio of the position of the cyclist in the lane to the total road space required to create the lane, i.e. the lane width added to the width of the separation. The higher ratio achieved by the wands is because they require less space in the road compared to other methods and cyclists use more of that space. Furthermore, standard specifications for kerb separation state a hard-margin of 0.5 metres, an additional m compared to the configuration tested in these trials, so the space saving on live roads would be even greater. Table 4 Road space required with each physical separation method and amount of space used by cyclists. Physical separation method Width of separation (m) Total space required by separation (with 2 m cycle lane) Average distance between cyclists and road edge (m) Separation efficiency ratio (Position of cyclist / Total space required by separation) Wands Zebras Kerb See Bikeability Delivery Guide, DfT April PPR704

24 Lateral position data show that cyclists made better use of the cycle lane width with the wand separation. This supports the finding that cyclists felt safest when using the cycle lane separated by wands, followed by the cycle lane separated by a solid kerb, and least safe when using the cycle lane separated by Zebras. The wands require less space for installation on the road compared to other methods and evidence suggests that they encourage cyclists to ride further out in the lanes, meaning they make greater use of the space that has been re-allocated to them. 2.4 What impact did vehicle interaction have on cyclist usability? Effect of vehicle interaction on ratings of ease of using the cycle lane On each trial day, cyclists were exposed to four vehicle interaction conditions on each cycle lane (i.e. the white line condition and one of the three physical separations). The four types of interaction were: no vehicle interaction, car passing cyclist at entrance to cycle lane, in middle of cycle lane and at exit of cycle lane. When cycling along the lane separated by Zebras, cyclists ratings of ease of using the cycle lane were marginally higher with no vehicle interaction compared to when a vehicle passed them at the exit. However, no other specific patterns in the on-track questionnaire responses were found. This suggests that the cyclists on-track ratings of usability were not greatly affected by the presence of a vehicle, regardless of the type of cycle lane separation. Ratings of usability were also obtained for three different traffic conditions using the offtrack questionnaire: no cars, one car and busy town centre traffic. The latter of these options required cyclists to imagine using the cycle lane in a hypothetical scenario, since busy town centre traffic was not trialled on the track. Figure 7 shows the mean off-track usability ratings for the white line separation and each of the methods of physical separation. It should be noted that the mean ratings across different days should not be directly compared because different groups of participants were tested on each of the 3 days. When comparing the ratings within each trial day, it is clear from that the ease of use was dependent on the level of traffic present when using the cycle lane, for all methods of separation. Specifically, ratings of ease of using the cycle lane decreased with increasing number of vehicles. April PPR704

25 Mean rating of ease of using cycle lane (off-track questionnaire) No traffic 1 car Busy traffic White line Wands White line Ziclas White line Kerb Day 1 Day 2 Day 3 Trial day Figure 7 Mean off-track questionnaire ratings for ease of using the cycle lane. Significant differences in off-track questionnaire ratings indicated an effect of traffic condition on the ease of using the separated cycle lane, regardless of the method of separation employed. Generally, ratings were lowest when cyclists were asked to imagine using the lanes in considered scenarios of increasing traffic volume from no cars, to one car, to busy traffic. 2.5 What impact did the vehicle interaction have on cyclist safety? Effect of vehicle interaction on ratings of perceived safety The type of vehicle interaction was also found to have a significant effect on the perceived safety of cyclists, regardless of the method of cycle lane separation. As expected, cyclists generally rated their journey as more safe when there was no vehicle interaction than when a car passed them at some point along the cycle lane. Mean safety ratings when a vehicle passed the cyclist in the middle of the cycle lane were typically marginally higher than when a vehicle passed the cyclist at either the entrance or the exit of the cycle lane. Differences were small but significant, suggesting that cyclists felt slightly more vulnerable to traffic both before joining the cycle lane and when re-joining the main carriageway. The magnitude of those differences was smaller for the physical separation methods than the non-physical separation, suggesting that the reduction in perceived safety associated with a vehicle interaction was mediated by the presence of physical separation. These findings are supported by responses to the off-track questionnaire, which obtained further ratings of safety for each cycle lane under three different traffic conditions: no cars, one car and busy town centre traffic. Figure 6 shows the mean off-track safety ratings for the white line separation and each of the physical methods of separation. April PPR704